INDEX

Explanations

personal names next to conjunctions

oai_token-act-pair · gpt-3.5-turbo

references to authors, editors, and other contributors in contexts related to published works

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Reviewer

-0.68

âĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢ

-0.65

eno

-0.62

TOR

-0.62

successfully

-0.61

enos

-0.60

DEN

-0.59

BTC

-0.59

enforcement

-0.59

istor

-0.58

POSITIVE LOGITS

 Seasons

0.58

 fourth

0.57

 Seas

0.52

 Susan

0.51

 sixth

0.51

 Edition

0.51

 Irma

0.50

 twins

0.50

 Epidem

0.50

 fifth

0.49

Activations Density 0.565%

personal names next to conjunctions

references to authors, editors, and other contributors in contexts related to published works

No Comments

No Known Activations